Git-based
Every dataset is an ordinary Git repository. Branches, tags, commits, and familiar tooling remain available.
A self-hosted data forge built on Git, Git LFS, and S3-compatible storage for researchers and machine-learning teams.

Niyān adds the coordination that large datasets need without hiding the tools underneath. Use the recommended CLI for the smoothest experience, or work with standard Git and Git LFS if you prefer.
$ pipx install niyan $ niyan auth login https://data.example.org $ niyan dataset clone lab/cell-images $ cd cell-images $ niyan status
Niyān is built on top of a few reliable components: Git records history, Git LFS represents large files, S3-compatible storage holds their contents, and Niyān's custom control plane manages identity, access, and the web experience.
Pin an immutable dataset commit beside your research code and return to exactly the same inputs later.
Clone, inspect, and recover datasets with established tools, even when the Niyān CLI is not available.
A focused control plane and a small number of durable services keep institutional deployment straightforward.